Frat Oncoproteins Act at the Crossroad of Canonical and Noncanonical Wnt-Signaling Pathways

Oncogene (2010) 29, 93–104 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 $32.00 www.nature.com/onc ORIGINAL ARTICLE Frat oncoproteins act at the crossroad of canonical and noncanonical Wnt-signaling pathways

R van Amerongen1,2,5, MC Nawijn1,2,6, J-P Lambooij1,2, N Proost1,2, J Jonkers3 and A Berns1,2,4

1Division of Molecular Genetics, Netherlands Cancer Institute, Amsterdam, The Netherlands; 2Centre of Biomedical Genetics, Amsterdam, The Netherlands; 3Division of Molecular Biology, Amsterdam, The Netherlands and 4Academic Medical Center, Amsterdam, The Netherlands

Wnt-signal transduction is critical for development and out the animal kingdom. Although Wnt proteins can tissue homeostasis in a wide range of animal species and is elicit multiple intracellular responses, activation of the frequently deregulated in human cancers. Members of the so-called ‘canonical’ or ‘Wnt/b-catenin’ pathway is Frat/GBP family of glycogen synthase kinase 3b (Gsk3b)- currently best understood. In the absence of extra- binding oncoproteins are recognized as potent activators of cellular Wnt, free cytoplasmic b-catenin is sequeste- the Wnt/b-catenin pathway in vertebrates. Here, we reveal red by a multiprotein complex containing APC, Axin1, a novel, Gsk3b-independent function of Frat converging casein kinase I, and glycogen synthase kinase 3b on the activation of JNK and AP-1. Both these have (Gsk3b). Sequential phosphorylation of b-catenin by been used as readouts for the noncanonical Frizzled/PCP the latter two kinases ensures its rapid degradation by pathway, which controls polarized cell movements and the the proteasome (Figure 1a). Binding of Wnt to the establishment of tissue polarity. We find that Frat Frizzled/LRP transmembrane receptor complex results synergizes with Diversin, the mammalian homolog of the in inhibition of Axin and Gsk3b, allowing the accumu- Drosophila PCP protein diego, in the activation of JNK/ lation of transcriptionally active b-catenin/TCF com- AP-1 signaling. Importantly, Frat mutants deficient for plexes (Figure 1b). The initiation of Wnt/b-catenin binding to Gsk3b retain oncogenic activity in vivo, signaling remains an area of intense study, and although suggesting that Wnt/b-catenin-independent events contri- some of the initiating events at the cell membrane have bute to Frat-induced malignant transformation. The been revealed (Bilic et al., 2007; Zeng et al., 2008), the observed activities of Frat are reminiscent of the dual precise mechanism behind Gsk3b inhibition remains to function of Dishevelled in the Wnt/b-catenin and Frizzled/ be resolved. PCP pathways and suggest that Frat may also function to In vertebrates, members of the Frat/GBP family of bridge canonical and noncanonical Wnt pathways. Gsk3b-binding proteins have been shown to compete Oncogene (2010) 29, 93–104; doi:10.1038/onc.2009.310; with Axin for binding to Gsk3b (Yost et al., 1998; Li published online 5 October 2009 et al., 1999a; Ferkey and Kimelman, 2002; Fraser et al., 2002), thereby disrupting the destruction complex and Keywords: Frat; Gsk3-binding protein; lymphoma- causing the accumulation of b-catenin (Figure 1c). genesis; Diversin; Wnt-signal transduction; planar cell Whereas the amphibian Frat ortholog GBP is critically polarity required for axis formation in Xenopus as part of the maternal Wnt pathway (Yost et al., 1998), Frat triple- knockout mice are viable without any overt phenotypic aberrations, indicating that Frat is not essential for Introduction Wnt/b-catenin signaling in mammals (van Amerongen et al., 2005). However, in support of a function for Frat Wnt-signaling controls a variety of biological processes, in canonical Wnt-signal transduction, the endogenous including cell proliferation, cell-fate decisions, and Frat-expression pattern shows remarkable overlap with polarized cell movements. As such, it is crucial for known anatomical sites of active Wnt/b-catenin signal- development and tissue homeostasis in species through- ing (Maretto et al., 2003; van Amerongen et al., 2005). This, together with the fact that Frat remains one of the most potent activators of b-catenin/TCF signaling Correspondence: Dr A Berns, Department of Molecular Genetics, identified to date, suggests that Frat acts as a modifier Netherlands Cancer Institute, Plesmanlaan 121, 1066 CX Amsterdam, to amplify canonical Wnt-pathway activity only under The Netherlands. E-mail: [email protected] specific circumstances. 5Current address: Department of Developmental Biology, Stanford The founding member of the Frat/GBP family, Frat1, University, Stanford, CA 94305, USA. was originally identified as a proto-oncogene in 6Current address: Laboratory Allergology and Pulmonary Diseases, advanced murine T-cell lymphomas (Jonkers et al., Section Pathology and Medical Biology, Groningen University Medical Centre, 9713 GZ, Groningen, The Netherlands. 1997). FRAT overexpression has been observed in a Received 25 March 2009; revised 3 August 2009; accepted 30 limited number of human malignancies (Saitoh and August 2009; published online 5 October 2009 Katoh, 2001; Wang et al., 2006c). FRAT1 was recently Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 94 Wnt

Frizzled/LRP Frizzled/LRP Frizzled/LRP Frizzled

Frat/GBP Dvl Dvl

Axin/APC/Gsk3 Axin/APC/Gsk3 Axin/APC/Gsk3 JNK Axin/APC/Gsk3

β-catenin β-catenin β -catenin β-catenin β-catenin β-catenin β-catenin

β-catenin/TCF β-catenin/TCF β -catenin AP-1 TCF

TCF target genes TCF target genes target genes

Wnt/β-catenin OFF Wnt/β-catenin ON Frat/GBP mediated activation Dishevelled functions at the crossroad of of the Wnt/β-catenin pathway Wnt/β-catenin and Frizzled/PCP signaling through inhibition of Gsk3b

TOPFLASH luciferase reporter AP-1 luciferase reporter 40 9 8 C Δ myc 7 30 Frat1 Frat1 Frat2 Δ 6 Dvl2 Frat1Frat1 C Frat2 myc 5 20 4

10 2 Fold activity relative to vector 1

0 0 vector Dvl2 Frat1 Frat1ΔC Frat2 vector Dvl2 Frat1 Frat1ΔC Frat2

Figure 1 The function of Frat and Dishevelled in Wnt-pathway activation. (a) In the absence of extracellular Wnt, the levels of cytoplasmic b-catenin are controlled by a multiprotein complex containing Axin, APC, and Gsk3b. (b) Binding of Wnt proteins to the Frizzled/LRP receptor complex causes the inhibition of Axin and Gsk3b, allowing b-catenin to accumulate and interact with transcription factors of the TCF/LEF family. (c) Vertebrate Frat/GBP proteins, whereas not critically required for Wnt/b-catenin signaling, are potent inhibitors of Gsk3b and cause an increase in b-catenin/TCF-dependent transcription. (d) Dishevelled functions at the crossroad of Wnt/b-catenin and Frizzled/PCP pathways. It promotes signaling through b-catenin/TCF, but has also been shown to induce the activity of JNK and AP-1. (e) Increasing amounts (25, 50, and 100 ng of DNA transfected) of Dvl2, Frat1, and Frat2 are all able to induce Wnt/b-catenin signaling in 293T cells, as evidenced by the concentration-dependent activation of the b-catenin/TCF- responsive TOPFLASH luciferase reporter. Frat2 is the least potent in this assay. Inset is a western blot depicting the expression levels of Frat1, Frat1DC and Frat2 after detection with an antibody directed against the myc-tag. (f) Increasing amounts (50, 100, and 300 ng of DNA transfected) of Frat2 and Dvl2, but not Frat1, induce a dose-dependent response of an AP-1 luciferase reporter containing seven multimerized AP-1 consensus sites. This effect is not merely because of a difference in protein stability between Frat1 and Frat2, as Frat1DC, which accumulates to higher levels than Frat1 (van Amerongen et al., 2004), is also unable to activate AP-1-dependent transcription. Data are represented as mean±s.e. Inset is a western blot depicting the expression levels of Frat1, Frat1DC, and Frat2 after detection with an antibody directed against the myc-tag.

reported to be overexpressed in esophageal squamous is also required for the establishment of tissue polarity cell carcinomas, in which its expression levels were (Figure 1d; Wallingford et al., 2000; Hamblet et al., shown to correlate with the accumulation of b-catenin 2002; Wang et al., 2006a; Etheridge et al., 2008). Here, it (Wang et al., 2008). However, we have been unable functions as a critical component of a ‘noncanonical’ to ﬁnd direct evidence for increased signaling through Wnt pathway, hereafter referred to as Frizzled/PCP b-catenin/TCF in murine T-cell lymphomas with signaling, mediated through JNK and AP-1 (reviewed in an activated Frat1 allele (our unpublished results), Fanto and McNeill, 2004; Klein and Mlodzik, 2005; indicating that Frat might act on a divergent pathway Jones and Chen, 2007; Wang and Nathans, 2007). in this context. We hypothesized that Frat might also have a dual Frat/GBP has earlier also been shown to interact with function. Our data show that Frat activates JNK and Dishevelled (Yost et al., 1998; Li et al., 1999a; Fraser AP-1 in a Gsk3b-independent manner and to a similar et al., 2002). This observation has thus far been degree as Dvl. Of note, the Gsk3b-binding domain of explained by a model in which Frat bridges signaling Frat is dispensable for its oncogenic activity. We further from Dvl to Gsk3b in activation of the Wnt/b-catenin ﬁnd that Frat synergizes with Diversin, the mammalian pathway. However, Dishevelled has a dual function, and homolog of the Drosophila PCP protein diego,inthe in addition to activating the Wnt/b-catenin pathway, it activation of both b-catenin/TCF and JNK/AP-1 signaling.

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 95 As we also observe endogenous expression of Frat2 at phospho-speciﬁc antibody directed against the JNK anatomical sites displaying Frizzled/PCP-pathway activity, target residue, Ser63. In agreement with the observed we propose that Frat proteins function at the crossroad increase in JNK kinase activity, we detected elevated of Wnt/b-catenin and Frizzled/PCP-signaling pathways. levels of endogenous phospho-Ser63 c-Jun in cells This study does not only reveal an unexpected and novel overexpressing Dvl2 or Frat2 (Figure 2b and Supple- Gsk3b-independent activity for Frat/GBP proteins in mentary Figure 2b). Addition of an ATP-competitive oncogenesis, but also ﬁts with the notion that crosstalk small molecule JNK inhibitor (SP600125) prevented the between Wnt/b-catenin and Frizzled/PCP signaling repre- activation of JNK and restored phospho-Ser63 c-Jun sents an additional layer of regulation of Wnt-induced levels to wild type (Figures 2c and d; Supplementary signaling events. Figures 2f and g). Thus, Dvl2 and Frat2 activate the AP-1 protein c-Jun at least in part through JNK.

Results Frat1IAQA and Frat2IAQA are defective in b-catenin/TCF signaling Frat2 activates AP-1 Frat-mediated activation of b-catenin/TCF signaling Dishevelled homolog from Drosophila and Xenopus as results from its binding to and inhibition of Gsk3b. As well as murine Dvl1 and Dvl2 have all been shown to neither Gsk3b nor b-catenin/TCF has been implicated in activate JNK in mammalian NIH3T3 fibroblasts and Frizzled/PCP signaling, we hypothesized that the 293T cells (Boutros et al., 1998; Li et al., 1999b; Habas observed activation of JNK and AP-1 reflected a novel, et al., 2003). Consequently, JNK kinase assays have Gsk3b-independent activity of Frat. The Gsk3b-binding been used to readout Frizzled/PCP-pathway activity in domain of Frat has been mapped to a small, evolutio- mammalian cells. In addition, generic AP-1 luciferase nary conserved stretch of amino acids in the C-terminus constructs can be activated by a number of Frizzled/ (Figure 3a). Mutation of the C-terminal IKEA-box into PCP-pathway components, including Wnt5A (Hocevar IAQA (Figure 3b) has earlier been shown to disrupt et al., 2003) and Dvl (Li et al., 1999b; Hocevar et al., Gsk3b binding in the Xenopus Frat homolog GBP (Yost 2003; Cong et al., 2004). We, therefore, sought to et al., 1998). determine whether, similar to Dvl, Frat might be able to To allow discrimination between the Gsk3b-depen- activate JNK and AP-1 in mammalian cells. dent and -independent activities of Frat, we generated As shown earlier, Dvl2, Frat1, and Frat2 activate a expression constructs encoding mutant Frat proteins in Wnt/b-catenin-responsive TOPFLASH luciferase repor- which the IKEA sequence had been mutated into IAQA. ter in a concentration-dependent manner (Figure 1e). In Western blot analysis showed that mutant and wild-type agreement with our earlier observations, Frat2 is less proteins were expressed at similar levels (Figure 3i). We potent in inducing Wnt/b-catenin signaling than Frat1 next compared wild-type and mutant Frat side by side in (van Amerongen et al., 2004). In contrast, when we a TOPFLASH luciferase reporter assay for b-catenin/ compared Dvl2, Frat1, and Frat2 for their ability to TCF signaling. As expected, the activity of Frat1IAQA activate AP-1 luciferase reporter activity in these cells, toward Gsk3b was completely abolished. The activity of we found that both Dvl2 and Frat2, but not Frat1, Frat2IAQA was also markedly reduced, although it induced AP-1 activity in a dose-dependent manner retained up to 30% of its activity for reasons that are (Figure 1f and Supplementary Figure 1a). As Frat1DC, presently unknown (Figure 3c). A similar effect was which accumulates to higher levels than Frat1 (Figure 1e observed when we tested Frat and FratIAQA toward and f; van Amerongen et al., 2004), is also unable to inhibition of a second, independent Gs3kb target activate AP-1-dependent transcription, the observed (Supplementary Figure 3). Thus, a comparison of effect is unlikely to be solely because of a difference in wild-type Frat and FratIAQA allows discrimination protein levels between Frat1 and Frat2. between Gsk3b-dependent and -independent signaling events. Dvl2 and Frat2 induce endogenous JNK kinase activity and phosphorylation of c-Jun AP-1 activation by Frat2 is Gsk3b independent Dvl2 has earlier been shown to activate JNK when both To test whether Frat-mediated activation of JNK and proteins were expressed exogenously (Li et al., 1999b). AP-1 was Gsk3b independent, we first tested the To test whether Dvl2 and Frat2 were able to activate induction of AP-1 luciferase reporter activity by Frat1 endogenous JNK, we performed a JNK kinase assay on or Frat2 in the presence of the pharmacological Gsk3b lysates from 293T cells overexpressing either Dvl2 or inhibitor SB-216763 (Figure 3d and Supplementary Frat2. We observed increased levels of endogenous JNK Figure 1b). Although SB-216763 induced TOPFLASH kinase activity in cells stimulated with anisomycin and in activity up to 70-fold, it did not significantly alter AP-1 cells transfected with Dvl2 or Frat2 compared with activity, either in the presence or absence of Frat. vector control cells (Figure 2a and Supplemen- Furthermore, the Gsk3b-binding deficient Frat2IAQA tary Figures 2a and b). In an attempt to more directly mutant was also able to induce the AP-1 luciferase link AP-1 reporter gene activation to the induction reporter (Figure 3e and Supplementary Figure 3c). In of JNK kinase activity, we probed the same lysates for fact, we found Frat2IAQA to be a more potent activator of the activation status of endogenous c-Jun using a AP-1-dependent transcription than Frat2, even though

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 96 it has lost most of its activity in Gsk3b-dependent sig- catenin pathway, whereas the total Frat2IAQA pool is naling events (compare Figure 3e with 3c). This might be available for activating JNK and AP-1. explained by the fact that wild-type Frat2 can be Although Frat1 and Frat1IAQA did not induce AP1 sequestered by Gsk3b, which draws it into the Wnt/b- luciferase reporter activity (Figure 3e and Supplemen-

anisomycin

vector vector vector Dvl2 Frat2 Dvl2 Frat2

pSer63 pSer63

GST-c-Jun c-Jun 8 5 7 6 4 5 3 4 3 2 2 Relative intensity Relative intensity 1 1 0 0 vectoranisomycin vect Dvl2 Frat2 vector Dvl2 Frat2 or

vector Dvl2 Frat2 vector Dvl2 Frat2 _ _ _ _ + _ + _ + SP600125 (20 μM) + + + SP600125 (20 μM)

pSer63 pSer63

GST-c-Jun c-Jun 5 5

4 4

3 3

2 2 Relative intensity 1 Relative intensity 1

0 _ _ _ 0 + + + SP600125 (20 μM) _ + _ + _ + SP600125 (20 μM) vector Dvl2 Frat2 vectorDvl2 Frat2 Figure 2 Frat2 activates endogenous c-Jun via JNK. (a) Dvl2 and Frat2 (750 ng of DNA transfected) induce endogenous JNK kinase activity in 293T cells, as measured by the in vitro phosphorylation of the JNK target residue Serine 63 on GST-c-jun. (b) Overexpression of Dvl2 and Frat2 in 293T cells results in a concomitant increase in the endogenous levels of phosphorylated c-Jun. (c) Phosphorylation of Serine 63 on c-Jun mediated by Dvl2 and Frat2 is dependent on endogenous JNK activity as it is abolished in the presence of the ATP-competitive JNK inhibitor SP600125. (d) Similarly, phosphorylation of endogenous c-Jun on Dvl2 and Frat2 overexpression is also prevented by the addition of SP600125. Bar diagrams represent quantiﬁcation of the depicted experiments.

Figure 3 AP1 activation by Frat2 is Gsk3b independent. (a) Frat proteins are evolutionary conserved across vertebrate species. All Frat/GBP orthologs studied to date show strong conservation of the C-terminal Gsk3b-binding domain. A short peptide (FRATtide) comprising amino-acids 188–226 from human FRAT1 is sufficient to inhibit GSK3 (Thomas et al., 1999; Bax et al., 2001). Mutation of the IKEA-box to IAQA has been shown to disrupt Gsk3b binding in Xenopus GBP (Yost et al., 1998). (b) Introduction of an identical mutation in mouse Frat1 and Frat2 yields Frat1IAQA and Frat2IAQA.(c) A TOPFLASH luciferase reporter assay to monitor Wnt/b- catenin-pathway activation shows that Frat1IAQA and Frat2IAQA are either completely or largely impaired in their ability to activate b- catenin/TCF signaling through inhibition of Gsk3b in 293T cells. Data are represented as mean±s.e. (d) The pharmacological Gsk3b inhibitor SB-216763 induces activation of the b-catenin/TCF-responsive TOPFLASH promoter. In contrast, it does not affect activation of an AP-1-responsive promoter. Moreover, Frat2 (300 ng of DNA transfected) is still able to activate AP-1 in the presence of the inhibitor. (e) Frat2IAQA (400 ng of DNA transfected) is a more potent activator of AP1-dependent transcription than wild-type Frat2 (400 ng of DNA transfected). (f and g) Frat1 and Frat1IAQA (750 ng of DNA transfected) induces endogenous JNK kinase activity in 293T cells to a lesser extent than Frat2 and Frat2IAQA (750 ng of DNA transfected), as measured by the in vitro phosphorylation of the JNK target residue Serine 63 on GST-c-jun. Bar diagrams represent quantification of the depicted experiments. (h) Direct lysates from the experiment depicted in (g). Overexpression of Frat1 and Frat1IAQA in 293T cells causes an increase in the endogenous levels of phosphorylated c-Jun, but to a lesser extent than and Frat2IAQA. Bar diagram represents quantification of the depicted experiment. (i) Western blot from the experiment depicted in (g–i) showing the levels of myc-Frat, phosphorylated JNK, and total JNK.

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 97 tary Figures 1b and c), we did observe a slight increase Frat2, but not Frat1, is expressed at sites in JNK kinase activity and phosphorylated c-jun levels of Frizzled/PCP-pathway activity when the proteins were expressed in 293T cells Our in vitro experiments suggested a Gsk3b-independent (Figures 3f–i; Supplementary Figures 2f and g). activity that was more prominent for Frat2 than for

mouse Frat1 168 RNAAASRRLQQRRGSQPETRTGDDDDPHRLLQQLVLSGNLIKEAVRRLHSRQLQLHAKLP mouse Frat2 149 PGSVASHRIQQRRWTAGGARAAD-DDPHRLLQQLVLSGNLIKEAVRRLQR------AVA- mouse Frat3 168 RNAAASRRQQQRRGSQSETRTSD-GDPPRLLQQLLLSGNLIKEAVRRLHSRGLQLQAKLP human FRAT1 175 RGAAASRRLQQRRGSQPETRTGD-DDPHRLLQQLVLSGNLIKEAVRRLHSRRLQLRAKLP human FRAT2 151 RDAVTSRRLQQRRWTQAGARAGD-DDPHRLLQQLVLSGNLIKEAVRRLQR------AVA- rat Frat 175 RNTAASRRLQQRRGSHLETRTGE-DDPPRLLQQLLLSGNLIKEAVRRLHSRRLQLQAKLP Xenopus GBP 110 ------CNCRKHAG----TEEEDDPHELLQELLLSGNLIKEAVRRLHMAG------zebrafish GBP 98 –----FKSSWNRKKI------DEDDPHQLLQELILSGNLIKEAVRRLQFS------

FRATtide SQPETRTGD-DDPHRLLQQLVLSGNLIKEAVRRLHSRRLQ Gsk3b-binding domain mutant IAQA

TOPFLASH 40 IKEA Frat1 1 274 35 IAQA 30 Frat1IAQA 1 274 25

IKEA 20 Frat2 1 232 15 IAQA 10 Frat2IAQA 1 232 5

TOPFLASH AP-1 luciferase reporter Fold activity relative to vector alone 0 80 luciferase reporter 10 Frat1 lAQA Frat2 lAQA 70 vector 8 Frat1 Frat2 60 AP-1luciferase reporter 25 50 6 40 20 30 4

20 15 2 10 Fold activity relative to vector alone

Fold activity relative to vector alone 10 0 0 vector vector Frat1 Frat2 -+SB-216763 --++ -+ SB-216763 5 Fold activity relative to vector alone 0

lAQA lAQA vector Frat1 Frat2

Frat1 Frat2

lAQA lAQA lAQA lAQA lAQA lAQA

vector Frat1 Frat2 vector Frat1 Frat1 Frat2 Frat2 vector Frat1 Frat1 Frat2 Frat2 vector Frat1 Frat1 Frat2 Frat2

pSer63 pSer63 pSer63

kinase assay kinase assay lysates (c-Jun) 6 (GST-c-Jun) (GST-c-Jun) myc 3 3 5 4 2 2 3 pJNK 2 1 1 1 Relative intensity Relative intensity 0 Relative intensity 0 0 JNK Frat1 lAQA Frat2 lAQA Frat1 lAQA Frat2 lAQA vector Frat1 Frat2 vector vector Frat1 Frat2 Frat1 Frat2

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 98 Frat1 in the activation of JNK/AP-1. However, neither modiolus. In addition, Frat1 was also expressed in Frat1 nor Frat2 has been shown to be involved in nonsensory supporting cells adjacent to the neuroe- Frizzled/PCP signaling in vivo. We have earlier gener- pithelium (Figure 4a). Remarkably, Frat2 was speciﬁ- ated Frat-knockout mice in which the Frat1 or Frat2 cally expressed in the neurosensory cells of the organ of coding sequence was replaced by a promoterless lacZ Corti (Figure 4b). When the X-gal-stained cochleas were gene, allowing us to monitor the endogenous Frat- sectioned, Frat2 expression was conﬁrmed to be present expression pattern. To determine whether Frat is in both the inner and outer hair cells, which are the expressed at anatomical sites that are known to display cells responsive to Frizzled/PCP signaling (Figure 4c). active Frizzled/PCP –signaling, we isolated cochlea from Importantly, expression of Frat2 does not coincide with adult Frat1 þ /lacZ and Frat2 þ /lacZ mice and performed Wnt/b-catenin signaling in the murine cochlea, as wholemount X-gal staining to identify lacZ-expressing neither BATGAL nor Axin2-lacZ mice, two different cells (Figure 4). Whereas Frat1 and Frat2 are co- b-catenin/TCF reporter mice, express lacZ in inner ear expressed in a number of different tissues known to hair cells (Qian et al., 2007 and data not shown). Its display active Wnt/b-catenin signaling (van Amerongen expression pattern is, however, remarkably similar to et al., 2005), we found that Frat1 and Frat2 were that of Frizzled3 and Frizzled6 (Wang et al., 2006b), mutually exclusively expressed in the different cell types both of which are involved in the establishment of PCP of the murine cochlea. Both Frat1 and Frat2 were in the mouse cochlea. Analysis of the inner ear hair cell expressed in the spiral ganglion of the cochlear orientation in Frat triple-knockout mice failed to reveal

OHC IHC

P P SG SG D D D

HC HC

SG SG IHC

OHC1

OHC2 HC HC OHC3

Frat1+/-;Frat2+/-;Frat3+/-

SG SG

IHC SL SL

IS IS OHC1 IHC IHC OHC2 OHC OHC OHC3 Frat1-/-;Frat2-/-;Frat3-/-

Figure 4 Frat2 is expressed at sites of Frizzled/PCP-pathway activity. X-gal staining in Frat1 þ /lacZ and Frat2 þ /lacZ mice reveals complementary expression patterns of Frat1 and Frat2 in the murine cochlea. (a) Frat1 is expressed in spinal ganglia and nonsensory supporting cells. (b and c) Frat2 is expressed in the inner and outer hair cells of the neurosensory epithelium, otherwise known as the organ of Corti. (c) Expression of Frat is speciﬁcally observed in the hair cells and not in the underlying Pillar and Deiter support cells. (d and e) Analysis of the inner ear hair cell orientation in cochlear ﬂatmounts from Frat1;Frat2;Frat3 triple-heterozygote or triple- knockout mice does not reveal any PCP defects. Abbreviations of anatomical landmarks: HC, hair cells; SG, spinal ganglion; SL, spiral limbus; IS, inner sulcus; IHC, inner ear hair cells; OHC, outer ear hair cells; D, Deiter cells; P, Pillar cells.

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 99 any abnormalities, indicating that Frat is not critically the Frizzled/PCP pathway, we asked whether Frat1 and required for the establishment of PCP in this tissue Frat2 might also cooperate with Diversin in activation (Figures 4d and e). of AP-1 signaling. In line with published data (Schwarz-Romond et al., Frat1 and Frat2 synergize with Diversin in Wnt/b-catenin 2002), we observed strong synergy between Frat1 and and Frizzled/PCP signaling Diversin in activation of the TOPFLASH reporter. Frat1 has earlier been reported to synergize with the Moreover, we observed a similar synergy for Frat2. In ankyrin-repeat protein Diversin/Ankrd6 in activation of agreement with the model proposed by Schwarz- the Wnt/b-catenin pathway (Schwarz-Romond et al., Romond et al., this synergy requires binding of Frat 2002). As both vertebrate Diversin as well as its to Gsk3b, as both Frat1IAQA and Frat2IAQA no longer Drosophila homolog diego have been shown to have a synergize with Diversin in TOPFLASH induction function in the establishment of tissue polarity as part of (Figure 5c; Supplementary Figures 1d and e). In contrast,

IAQA IAQA

vector DiversinFrat1 Frat1 Frat2 Frat2 Diversin +Frat1Diversin+Frat1DiversinDiversin + Frat2 +Frat2

ANK CK CBD Diversin flag ankyrin repeats casein kinase conductin/Axin2 binding domain binding domain

Frat myc

80 TOPFLASH luciferase reporter 10 AP-1 luciferase reporter

9 70 8 60 7

50 6

40 5

4 30 3 20 2 Fold activity relative to vector alone Fold activity relative to vector alone 10 1

0 0 vector Frat1 Frat1IAQA Frat2 Frat2IAQA vector Frat1 Frat1IAQA Frat2 Frat2IAQA Diversin - + - + - + -+ -+ Diversin - + - + - + -+ - +

35 TOPFLASH luciferase reporter AP-1 luciferase reporter 10 flag 30 9 8 DIVV 25 7 DIV-ΔCBD

20 6 5 15 4

10 3 2

Fold activity relative to vector alone 5 Fold activity relative to vector alone 1 DIV-CBD 0 0 vector DIV DIV-CBD DIV-ΔCBD vector DIV DIV-CBD DIV-ΔCBD DIV DIV-CBDDIV- Δ Frat2 - + - + - + - + Frat2 - +-+ - + - + CBD

Figure 5 Synergy between Frat and Diversin in Wnt/b-catenin and Frizzled/PCP signaling. (a) Schematic representation of the different functional domains in Diversin. (b) Western blot depicting the expression levels of FLAG-Diversin and myc-Frat. (c) In 293T cells, Diversin synergizes with Frat1 and Frat2, but not Frat1IAQA and Frat2IAQA in activation of the canonical Wnt pathway as measured by induction of TOPFLASH luciferase reporter activity. (d) In contrast, Diversin synergizes with Frat1 and Frat2 in a Gsk3b-independent manner in activation of an AP-1 luciferase reporter. (e) The Axin2/CBD of Diversin is both necessary and sufﬁcient for the synergizing effect observed with Frat in activation of the canonical Wnt pathway (100 ng Frat2; 100 ng Diversin DNA transfected). (f) In contrast, deletion of the Diversin CBD does not abolish synergy between Frat and Diversin in activation of the noncanonical Wnt pathway (150 ng Frat2; 300 ng Diversin DNA transfected). (g) Western blot depicting the expression levels of the different FLAG-tagged Diversin constructs.

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 100 we observe that Diversin synergizes with both wild-type have not yet been shown to result in an increase in Frat2 Frat2 and Frat2IAQA in the activation of AP-1, under- levels. Likewise, the original tumor panel in which Frat1 scoring that this is a separate, Gsk3b-independent was first identified contained a limited number of tumors activity (Figure 5d; Supplementary Figures 1d and e). that appeared to harbor proviral insertions in the Frat2 Interestingly, Frat1 and Frat1IAQA were also able to promoter. However, we were unable to detect a fusion activate the AP-1 reporter in the presence of Diversin, transcript or an increase in Frat2 expression levels in albeit less efficient than Frat2. these tumors (our unpublished results). Analogous to To further investigate the mechanism behind the the situation observed for the Pim proto-oncogenes (van observed synergy between Frat and Diversin, we took der Lugt et al., 1995; Mikkers et al., 2002), Frat2 might advantage of a series of Diversin deletion mutants only become a predominant target for M-MuLV generated by Moeller et al. (2006) (Figure 5a)), which insertions in a Frat1-knockout background, in which allowed us to test the requirement of the Axin2/ case it might complement the loss of Frat1. To test this conduction-binding domain (CBD) of Diversin for the hypothesis, we injected newborn mice that were either observed synergy with Frat in activation of the two wild type (Frat1 þ / þ ), or had lost one (Frat1 þ /À) or both branches of the Wnt pathway. (Frat1À/À) alleles of Frat1 with M-MuLV and monitored Synergy between Frat2 and Diversin in activation of lymphoma incidence (Supplementary Figure 4). Although the Wnt/b-catenin pathway was solely dependent on the tumor formation was slightly delayed in Frat1 þ /À and Diversin CBD domain, as evidenced by the fact that Frat1À/À mice, Southern blot analysis did not reveal any DIV-CBD, expressing only the Axin2-binding domain, retroviral insertions in the vicinity of Frat2. Thus, loss of was capable of enhancing TOPFLASH induction by Frat1 does not substantially increase the selective Frat2, whereas all synergy was lost in the DIV-DCBD advantage of cells carrying a provirally activated Frat2 mutant lacking only the Axin2-binding domain in primary lymphomas. (Figure 5e and Supplementary Figure 1f). In contrast, We earlier observed that overexpression of Frat1 synergy between Frat2 and Diversin in Frizzled/PCP- enhances the sensitivity to tumorigenesis as illustrated pathway activation was only observed for full-length by the fact that Frat1/Pim1 double-transgenic mice Diversin and DIV-DCBD, suggesting that this effect show a higher incidence of spontaneous lymphomas occurred independent from Axin (Figure 5f and compared with Pim1-transgenic mice alone (Jonkers Supplementary Figure 1f). Thus, Frat synergizes with et al., 1999b). To determine the function of Frat2 as Diversin in Wnt/b-catenin signaling in an Axin- and well as the Gsk3b-binding domain of Frat in tumor Gsk3b-dependent manner. In contrast, Frat and Diver- formation, we performed bone marrow transplantation sin promote JNK/AP-1 signaling independent from assays of Pim1-transgenic bone marrow transduced with Axin and Gsk3b. retroviral vectors encoding either wild-type Frat1 or Frat2, or the Gsk3b-binding mutants Frat1IAQA and Frat2IAQA. Figure 6 shows the rate of tumor onset after Frat1 and Frat2-driven T-cell lymphomagenesis transplantation of the infected bone marrow into in mice occurs independent of Gsk3b irradiated recipients. Mice that received Pim1-transgenic Frat1 was originally identified in a retroviral insertional bone marrow transduced with either Frat1 or Frat2 mutagenesis screen designed to uncover genes that were developed lymphomas with faster kinetics than controls, specifically involved in later stages of tumor progression. which received Pim1-transgenic bone marrow trans- As such, activation of Frat1 was predominantly duced with empty vector. Thus, Frat2 functions as an observed in advanced disease after transplantation of oncogene and, similar to Frat1, synergizes with Pim1 in M-MuLV-induced murine T-cell lymphomas (Jonkers lymphomagenesis. Remarkably, mice that received et al., 1997). Whereas we did observe an association Pim1-transgenic bone marrow transduced with either between Frat1 and Gsk3b in the T-cell lymphomas in Frat1IAQA or Frat2IAQA also showed an increase in which Frat1 was first identified, we have so far been lymphomagenesis compared with controls, indicating unable to obtain conclusive evidence showing increased that at least part of the oncogenic activities of Frat1 and b-catenin/TCF signaling in these tumors (data not Frat2 proceed independent from Gsk3b and Wnt/b- shown). catenin signaling. Given that Frat1 is a more potent activator of the Wnt/b-catenin pathway than Frat2 (van Amerongen et al., 2004), we asked whether Frat2 could also promote tumorigenesis. In light of the newfound Gsk3b-inde- Discussion pendent activity of Frat toward JNK and AP-1, we also sought to determine whether the Gsk3b-dependent Frat/GBP proteins are potent activators of b-catenin/ or -independent activities contribute to the function of TCF signaling. Efforts determining to what extent Frat Frat in tumor progression. might influence other cellular processes have remained At present, a direct function in oncogenesis has only limited. Here, we show for the first time that Frat has been shown for Frat1. High throughput M-MuLV biological activities other than inducing b-catenin/TCF insertional mutagenesis screens have revealed proviral signaling, and we provide evidence for a Gs3kb- integrations that might cause activation of Frat2 independent function of Frat in the activation of (J Kool, personal communication), but these insertions JNK and AP-1. Of note, Frat mutants deficient in

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 101 Pim1-transgeniccontrol irradiated C57/B6J recipient Frat1 Frat1IAQA Frat2 Frat2IAQA

100 100 Control 80 80 Frat2 Frat2IAQA 60 60

40 40 Control

Percent survival Frat1 20 Percent survival 20 Frat1IAQA 0 0 0 100 200 300 400 500 0 100 200 300 400 500 Days after grafting Days after grafting Figure 6 The oncogenic activities of Frat are Gsk3b independent. (a) Bone marrow was isolated from Pim1-transgenic mice. After infection with retroviruses encoding empty vector, myc-Frat1, myc-Frat1IAQA, myc-Frat2, or myc-Frat2IAQA, cells were transplanted into irradiated recipients, which were monitored for lymphoma development. (b) Compared with empty vector, both wild-type Frat1 (n ¼ 5; P ¼ 0.0494; log-rank (Mantel–Cox) test) and Frat2 (n ¼ 5; P ¼ 0.0221; log-rank (Mantel–Cox) test), as well as Frat1IAQA (n-5; P ¼ 0.0221; log-rank (Mantel–Cox) test) synergize with Pim1 in lymphomagenesis. Although Frat2IAQA (n ¼ 5; P ¼ 0.0888; log-rank (Mantel–Cox) test) failed to reach signiﬁcance, it showed a similar trend. Animals were followed over a period of 470 days. Mice that died without any discernable signs of lymphoid tumors were considered censored subjects.

Wnt/b-catenin signaling retain oncogenic activity. This, when Frat protein levels are elevated in response to a together with the fact that we find Frat2 to be expressed particular, as of yet unknown, signal. In fact, it is at sites of known Frizzled/PCP activity, suggests that precisely in one such setting that Frat1 was originally in vivo Frat functions at least in part independent of identified: stabilization of the Frat1 transcript after Gsk3b and b-catenin/TCF. insertion of an M-MuLV provirus allowed Frat1 to Both murine Frat1 and Xenopus GBP were earlier exert its oncogenic activities in advanced stages of T-cell shown to interact with Dvl. This interaction was initially lymphomagenesis. proposed to bridge transmission of a ‘canonical’ Wnt To test the relevance of the Gsk3b-independent effects signal from Dvl to Gsk3b (Li et al., 1999a). However, of Frat in vivo, we, therefore, turned to a cancer model Dvl is also a core PCP protein and as such it has been in which Frat1 has earlier been shown to provide a considered as the bifurcation point of canonical and competitive advantage to the tumor cells. Here, we noncanonical Wnt pathways (Theisen et al., 1994; found that not only wild-type Frat1 and Frat2, but also Boutros et al., 1998; Li et al., 1999b; Wallingford the Gsk3b-binding mutants Frat1IAQA and Frat2IAQA, are et al., 2000). The activation of both JNK and AP-1 has able to synergize with Pim1 in lymphomagenesis been used to read out Frizzled/PCP-pathway activation (Figure 6). This is the first demonstration in which at in mammalian cells. We show that Frat2, but not Frat1, least part of the oncogenic activities of Frat1 and Frat2 is able to induce AP-1 activity in 293T cells as efficiently are independent of Gsk3b and possibly mediated as Dvl2 in a dose-dependent manner (Figure 1 and through the Frizzled/PCP-pathway effectors JNK and Supplementary Figure 1). We further show that Frat2 AP-1, although at present the critical targets of Frat1 causes an increase in endogenous JNK kinase activity and/or Frat2 in lymphomagenesis remain unknown. and activated c-Jun levels (Figure 2). Importantly, these Although we observe a slight increase in JNK activity activities are Gsk3b independent (Figure 3 and Supple- induced by Frat1 and Frat1IAQA, this does not translate mentary Figure 1). into the induction of AP-1 signaling, except when Interestingly, we also find that Frat2, but not Frat1, is Diversin is overexpressed (Figures 3 and 5; Supplemen- expressed in inner ear hair cells (Figure 4). Analysis of tary Figure 2). Thus, Frat2 seems to be a more potent Frat triple-knockout mice has so far failed to expose any activator of JNK/AP-1 signaling than Frat1 or Frat1IAQA. phenotypic abnormalities in either Wnt/b-catenin or Although this difference can in part be explained by Frizzled/PCP signaling, suggesting that cells do not the difference in protein levels between Frat1 and Frat2 critically require Frat under steady state conditions. As (Figures 1e, f and 3i; van Amerongen et al., 2004), it is a result, the effects of Frat deficiency might only be also likely caused by intrinsic differences between Frat1 revealed in a predisposed background. Alternatively, and Frat2 that are currently not understood. Moreover, given that the endogenous Frat transcripts and proteins we clearly observe acceleration of lymphomagenesis by are short lived and expressed at borderline detection Frat1IAQA in vivo, suggesting that in the hematopoietic levels, the function of Frat might only become apparent system, the conditions are such that Frat1IAQA can also

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 102 Wnt/Frizzled Materials and methods

? ? DNA constructs Frat1 Dvl Frat2 (and Frat2) (and Frat1) Generation of myc-tagged expression constructs encoding full- Diversin Diversin length Frat1 and Frat2 or a Frat1 C-terminal deletion construct CKIε CKIε has been described before (van Amerongen et al., 2004). IAQA IAQA Axin/APC/Gsk3 JNK Frat1 and Frat2 were constructed by site-directed mutagenesis of myc-Frat1 and myc-Frat2. For retroviral infections, myc-Frat1, myc-Frat1IAQA, myc-Frat2, and myc- Frat2IAQA wereclonedintoMSCV-IRES2-GFP,whichwas β-catenin/TCF c-jun generated by cloning the IRES2 from pIRES2-DsRed (Clontech, Mountain View, CA, USA) into MSCV-IRES-GFP. TOPFLASH was a gift from Dr Hans Clevers (Utrecht, TCF target genes AP-1 target genes The Netherlands), HA-Dvl2 was a gift from Dr Trevor Dale (Cardiff, UK), Diversin constructs were a gift from Dr Wnt/β-catenin Frizzled/PCP Walter Birchmeier (Berlin, Germany), and GFP-NFATc1 was Figure 7 Frat and Diversin function at the crossroad of Wnt/b- a gift from Dr Gerald Crabtree (Stanford, California, CA, catenin and Frizzled/PCP signaling. Model for the activites of USA). Additional constructs used were the AP-1 luciferase Frat1 and Frat2 in Wnt-signal transduction. Depending on the reporter (Stratagene, La Jolla, CA, USA), CMV-bgal (Clon- cellular context, Frat1 and Frat2 may preferentially amplify Wnt/ tech), and CMV-Renilla (Promega, Madison, WI, USA). b-catenin or Frizzled/PCP signaling (see text for details).

Animal experiments induce a Gsk3b-independent response. Interestingly, All animal experiments were performed conform national Frat1 was earlier described to inhibit AP-1 signaling regulatory standards approved by the DEC (Animal Experi- downstream of JNK by inhibiting the Gsk3b-dependent ments Committee). Generation of Frat1- and Frat2-knockout phosphorylation of c-Jun in neuronal cells (Hongisto mice has been described before (Jonkers et al., 1999a; van et al., 2003). This observation is not in conflict with our Amerongen et al., 2005). A detailed description can be found data, which show a Gsk3b-independent function for in the Supplementary Materials and methods. Frat in JNK/AP-1 signaling, but it does underscore once more that the function of Frat is likely to be context Cell culture and transfections dependent. 293T cells were grown in DMEM supplemented with 10% On our combined observations, we propose a model fetal bovine serum and 1% penicillin/streptomycin (Gibco, in which Frat functions at the crossroad of Wnt/b- Carlsbad, CA, USA) under 5% CO2 at 37 1C in humidifying catenin and Frizzled/PCP pathways (Figure 7). conditions. On the day before transfection, cells were plated in Although Frat1 and Frat2 are each able to activate 6-well plates (for kinase assays) or in 12-well tissue culture both pathways depending on the cellular context, they plates (for luciferase assays). Cells were transfected with a total amount of 1500 ng (for 6-well plates) or 500 ng (for 12-well do so with different affinities. Whereas Frat1 is more plates) DNA per well using polyethylenimine (PEI, Poly- likely to sensitize cells toward the Gsk3b-mediated sciences Inc., Warrington, PA, USA). In all cases, empty activation of b-catenin/TCF signaling, Frat2 preferen- pGlomyc was added to control for the total amount of DNA. tially causes the Gsk3b-independent activation of JNK Cells were treated with the JNK inhibitor SP600125 (20 mM and AP-1. Although other proteins, such as Naked (Yan stock in DMSO, SuperArray, Frederick, MD, USA) 24 h et al., 2001), Diversin (Schwarz-Romond et al., 2002), post-transfection and harvested another 24 h later. Protein and Inversin (Simons et al., 2005) also function as a concentration was determined using a colorimetric assay (Bio- switch between canonical and noncanonical Wnt path- Rad, Hercules, CA, USA). ways, they are usually reported to promote one branch, whereas inhibiting the other. Notable exceptions are Dvl Luciferase assays and casein kinase Ie ((Sakanaka et al., 2000; Klein et al., 293T cells were transfected as described above with 50–100 ng 2006; Strutt et al., 2006), which perform a dual function of TOPFLASH or AP-1 luciferase reporter, 100 ng of CMV- in the activation of both b-catenin/TCF and Frizzled/ bgal, or 2 ng of Renilla and the indicated amounts of empty PCP signaling. Interestingly, casein kinase Ie was earlier vector or various expression constructs. Cells were harvested shown to enhance the interaction between Dvl and Frat 48 h post-transfection in reporter lysis buffer or passive lysis (Hino et al., 2003). Furthermore, casein kinase Ie also buffer (Promega) and analyzed with luciferase assay reagent (Promega) according to the manufacturer’s instructions in a binds to Diversin (Supplementary Figure 5). It is TopCounter (Packard, Meriden, CT, USA) or Berthold Lumat tempting to speculate that these proteins might synergize as described in the Supplementary Materials and methods. in both b-catenin-dependent and -independent signaling events. In this respect, the dual activities of Frat are likely to underlie a more general, dynamic interaction Kinase assay between Wnt/b-catenin and Frizzled/PCP-pathway 293T cells were transfected as described above with the indicated constructs. Cells were harvested 48 h post-transfec- components. This notion is supported by the fact that tion in cell lysis buffer (Cell Signaling, Danvers, MA, USA) Diversin strongly promotes signaling through b-catenin/ and a kinase assay was performed using the JNK/SAPK TCF in collaboration with Frat. Such crosstalk would kinase assay kit (Cell Signaling) according to the manufac- provide cells with the flexibility and means to properly turer’s instructions (see Supplementary Materials and methods integrate upstream stimuli. for a more detailed description).

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 103 Western blot analysis Software Samples were run on 10% NuPage precast gels (Invitrogen, For bone marrow transplantation experiments, generation of Carlsbad, CA, USA), transferred to Protran membranes survival curves and statistical analyses were performed using (Schleicher and Schuell Bioscience Gmbh, Dassel, Germany), GraphPad Prism 5. Confocal images acquired on a Leica CSLM and analyzed using ECL (Pierce, Rockford, IL, USA). microscope were analyzed using CellProfiler software (Carpenter Antibodies were used recognizing the myc-tag (9E10, 1:5000, et al., 2006; Lamprecht et al., 2007) (available from www.cellpro- Invitrogen), the HA-tag (3F10, 1:2000, Roche, Indianapolis, filer.org). Nuclei were identified by means of the TOPRO3 signal, IN, USA), the FLAG-tag (M2, 1:2000, Sigma), pSer63-c-jun after which the total GFP signal was measured in the nuclei and (1:1000, Cell Signaling), pJNK (1:1000, Cell Signaling), JNK cytoplasm using an optimized pipeline. For quantification of (1:1000 Cell Signaling), c-jun (H79, 1:1000, Santa Cruz, Santa protein levels on western blots densitometry was performed using Cruz, CA, USA), and actin (C4, 1:1000, Sigma). Secondary Image J software (http://rsb.info.nih.gov/ij/). antibodies were goat-anti-mouse-HRP (1:5000, Biosource, Carlsbad, CA, USA), rabbit-anti-rat-HRP (1:2000, Dako, Glostrup, Denmark) and anti-rabbit-HRP (1:2000, Cell Conflict of interest Signaling). The authors declare no conflict of interest.

X-gal staining of murine cochlea Cochlea were isolated from adult Frat1 þ /À or Frat2 þ /À mice, fixed Acknowledgements overnight at 4 1C in PBS with 0.2% glutaraldehyde, EDTA, and MgCl2 and processed for X-Gal staining as described before (van We thank Dr Walter Birchmeier, Dr Hans Clevers, Dr Gerald Amerongen and Berns, 2005). After staining, cochleas were Crabtree, Dr Trevor Dale, Dr Jolita Hendriksen, Dr Rob dissected to expose the organ of Corti. Images were acquired Michalides, and Dr Rob Wolthuis for providing constructs and using an Olympus stereomicroscope and an Olympus camera. reagents and Dr Ping Chen for advice on cochlear dissections. Wholemount-stained cochlea were flatmounted and photo- We thank Fina van de Ahe´, Sjaak Greven, Loes Rijswijk, and graphed on a light microscope. Alternatively, wholemount- Maaike Voetel for tail vein injections, Hanneke van de Gulden stained cochlea were post-fixed in 4% paraformaldehyde, for generating the Frat1IAQA mutant, and Dr Joaquim Calbo- embedded, sectioned, and counterstained with nuclear fast red. Angrill for reading of an early version of the manuscript. This See Supplementary Materials and methods for analysis of inner work received financial support from the Centre of Biomedical ear hair cell orientation by confocal microscopy. Genetics (CBG).

References

Bax B, Carter PS, Lewis C, Guy AR, Bridges A, Tanner R et al. (2001). Hamblet NS, Lijam N, Ruiz-Lozano P, Wang J, Yang Y, Luo Z et al. The structure of phosphorylated GSK-3beta complexed with a (2002). Dishevelled 2 is essential for cardiac outflow tract peptide, FRATtide, that inhibits beta-catenin phosphorylation. development, somite segmentation and neural tube closure. Devel- Structure (Camb) 9: 1143–1152. opment 129: 5827–5838. Bilic J, Huang YL, Davidson G, Zimmermann T, Cruciat CM, Hino S, Michiue T, Asashima M, Kikuchi A. (2003). Casein kinase I Bienz M et al. (2007). Wnt induces LRP6 signalosomes and epsilon enhances the binding of Dvl-1 to Frat-1 and is essential for promotes dishevelled-dependent LRP6 phosphorylation. Science Wnt-3a-induced accumulation of beta-catenin. J Biol Chem 278: 316: 1619–1622. 14066–14073. Boutros M, Paricio N, Strutt DI, Mlodzik M. (1998). Dishevelled Hocevar BA, Mou F, Rennolds JL, Morris SM, Cooper JA, Howe PH. activates JNK and discriminates between JNK pathways in planar (2003). Regulation of the Wnt signaling pathway by disabled-2 polarity and wingless signaling. Cell 94: 109–118. (Dab2). EMBO J 22: 3084–3094. Carpenter AE, Jones TR, Lamprecht MR, Clarke C, Kang IH, Friman Hongisto V, Smeds N, Brecht S, Herdegen T, Courtney MJ, Coffey O et al. (2006). CellProfiler: image analysis software for identifying ET. (2003). Lithium blocks the c-Jun stress response and protects and quantifying cell phenotypes. Genome Biol 7: R100. neurons via its action on glycogen synthase kinase 3. Mol Cell Biol Cong F, Schweizer L, Varmus H. (2004). Casein kinase Iepsilon 23: 6027–6036. modulates the signaling specificities of dishevelled. Mol Cell Biol 24: Jones C, Chen P. (2007). Planar cell polarity signaling in vertebrates. 2000–2011. Bioessays 29: 120–132. Etheridge SL, Ray S, Li S, Hamblet NS, Lijam N, Tsang M et al. Jonkers J, Korswagen HC, Acton D, Breuer M, Berns A. (1997). (2008). Murine dishevelled 3 functions in redundant pathways with Activation of a novel proto-oncogene, Frat1, contributes dishevelled 1 and 2 in normal cardiac outflow tract, cochlea, and to progression of mouse T-cell lymphomas. EMBO J 16: neural tube development. PLoS Genet 4: e1000259. 441–450. Fanto M, McNeill H. (2004). Planar polarity from flies to vertebrates. Jonkers J, van Amerongen R, van der Valk M, Robanus-Maandag E, J Cell Sci 117: 527–533. Molenaar M, Destree O et al. (1999a). in vivo analysis of Frat1 Ferkey DM, Kimelman D. (2002). Glycogen synthase kinase-3 beta deficiency suggests compensatory activity of Frat3. Mech Dev 88: mutagenesis identifies a common binding domain for GBP and 183–194. Axin. J Biol Chem 277: 16147–16152. Jonkers J, Weening JJ, van der Valk M, Bobeldijk R, Berns A. (1999b). Fraser E, Young N, Dajani R, Franca-Koh J, Ryves J, Williams RS Overexpression of Frat1 in transgenic mice leads to glomerulo- et al. (2002). Identification of the Axin and Frat binding sclerosis and nephrotic syndrome, and provides direct evidence for region of glycogen synthase kinase-3. J Biol Chem 277: the involvement of Frat1 in lymphoma progression. Oncogene 18: 2176–2185. 5982–5990. Habas R, Dawid IB, He X. (2003). Coactivation of Rac and Rho by Klein TJ, Jenny A, Djiane A, Mlodzik M. (2006). CKIepsilon/discs Wnt/Frizzled signaling is required for vertebrate gastrulation. Genes overgrown promotes both Wnt-Fz/beta-catenin and Fz/PCP Dev 17: 295–309. signaling in Drosophila. Curr Biol 16: 1337–1343.

Oncogene Frat activates JNK/AP-1 independent from GSK3 R van Amerongen et al 104 Klein TJ, Mlodzik M. (2005). Planar cell polarization: an emerging Thomas GM, Frame S, Goedert M, Nathke I, Polakis P, Cohen P. model points in the right direction. Annu Rev Cell Dev Biol 21: 155–176. (1999). A GSK3-binding peptide from FRAT1 selectively inhibits Lamprecht MR, Sabatini DM, Carpenter AE. (2007). CellProﬁler: the GSK3-catalysed phosphorylation of axin and beta-catenin. free, versatile software for automated biological image analysis. FEBS Lett 458: 247–251. Biotechniques 42: 71–75. van Amerongen R, Berns A. (2005). Re-evaluating the role of Frat in Li L, Yuan H, Weaver CD, Mao J, Farr III GH, Sussman DJ et al. Wnt-signal transduction. Cell Cycle 4: 1065–1072. (1999a). Axin and Frat1 interact with dvl and GSK, bridging van Amerongen R, Nawijn M, Franca-Koh J, Zevenhoven J, van der Dvl to GSK in Wnt- mediated regulation of LEF-1. EMBO J 18: Gulden H, Jonkers J et al. (2005). Frat is dispensable for canonical 4233–4240. Wnt signaling in mammals. Genes Dev 19: 425–430. Li L, Yuan H, Xie W, Mao J, Caruso AM, McMahon A et al. (1999b). van Amerongen R, van der Gulden H, Bleeker F, Jonkers J, Berns A. Dishevelled proteins lead to two signaling pathways. Regulation of (2004). Characterization and functional analysis of the murine Frat2 LEF-1 and c-Jun N-terminal kinase in mammalian cells. J Biol gene. J Biol Chem 279: 26967–26974. Chem 274: 129–134. van der Lugt NM, Domen J, Verhoeven E, Linders K, van der Gulden Maretto S, Cordenonsi M, Dupont S, Braghetta P, Broccoli V, Hassan H, Allen J et al. (1995). Proviral tagging in E mu-myc transgenic AB et al. (2003). Mapping Wnt/beta-catenin signaling during mouse mice lacking the Pim-1 proto-oncogene leads to compensatory development and in colorectal tumors. Proc Natl Acad Sci USA 100: activation of Pim-2. EMBO J 14: 2536–2544. 3299–3304. Wallingford JB, Rowning BA, Vogeli KM, Rothbacher U, Fraser SE, Mikkers H, Allen J, Knipscheer P, Romeijn L, Hart A, Vink E et al. Harland RM. (2000). Dishevelled controls cell polarity during (2002). High-throughput retroviral tagging to identify components Xenopus gastrulation. Nature 405: 81–85. of speciﬁc signaling pathways in cancer. Nat Genet 32: 153–159. Wang J, Hamblet NS, Mark S, Dickinson ME, Brinkman BC, Segil N Moeller H, Jenny A, Schaeffer HJ, Schwarz-Romond T, Mlodzik M, et al. (2006a). Dishevelled genes mediate a conserved mammalian Hammerschmidt M et al. (2006). Diversin regulates heart formation PCP pathway to regulate convergent extension during neurulation. and gastrulation movements in development. Proc Natl Acad Sci Development 133: 1767–1778. USA 103: 15900–15905. Wang Y, Guo N, Nathans J. (2006b). The role of Frizzled3 and Qian D, Jones C, Rzadzinska A, Mark S, Zhang X, Steel KP et al. Frizzled6 in neural tube closure and in the planar polarity of inner- (2007). Wnt5a functions in planar cell polarity regulation in mice. ear sensory hair cells. J Neurosci 26: 2147–2156. Dev Biol 306: 121–133. Wang Y, Hewitt SM, Liu S, Zhou X, Zhu H, Zhou C et al. (2006c). Saitoh T, Katoh M. (2001). FRAT1 and FRAT2, clustered in human Tissue microarray analysis of human FRAT1 expression and its chromosome 10q24.1 region, are up- regulated in gastric cancer. Int correlation with the subcellular localisation of beta-catenin in J Oncol 19: 311–315. ovarian tumours. Br J Cancer 94: 686–691. Sakanaka C, Sun TQ, Williams LT. (2000). New steps in the Wnt/beta- Wang Y, Liu S, Zhu H, Zhang W, Zhang G, Zhou X et al. (2008). catenin signal transduction pathway. Recent Prog Horm Res 55: FRAT1 overexpression leads to aberrant activation of beta-catenin/ 225–236. TCF pathway in esophageal squamous cell carcinoma. Int J Cancer Schwarz-Romond T, Asbrand C, Bakkers J, Kuhl M, Schaeffer HJ, 123: 561–568. Huelsken J et al. (2002). The ankyrin repeat protein Diversin Wang Y, Nathans J. (2007). Tissue/planar cell polarity in recruits Casein kinase Iepsilon to the beta-catenin degradation vertebrates: new insights and new questions. Development 134: complex and acts in both canonical Wnt and Wnt/JNK signaling. 647–658. Genes Dev 16: 2073–2084. Yan D, Wallingford JB, Sun TQ, Nelson AM, Sakanaka C, Reinhard Simons M, Gloy J, Ganner A, Bullerkotte A, Bashkurov M, Kronig C C et al. (2001). Cell autonomous regulation of multiple Dishevelled- et al. (2005). Inversin, the gene product mutated in nephronophthi- dependent pathways by mammalian Nkd. Proc Natl Acad Sci USA sis type II, functions as a molecular switch between Wnt signaling 98: 3802–3807. pathways. Nat Genet 37: 537–543. Yost C, Farr III GH, Pierce SB, Ferkey DM, Chen MM, Kimelman D. Strutt H, Price MA, Strutt D. (2006). Planar polarity is positively (1998). GBP, an inhibitor of GSK-3, is implicated in Xenopus regulated by casein kinase Iepsilon in Drosophila. Curr Biol 16: development and oncogenesis. Cell 93: 1031–1041. 1329–1336. Zeng X, Huang H, Tamai K, Zhang X, Harada Y, Yokota C et al. Theisen H, Purcell J, Bennett M, Kansagara D, Syed A, Marsh JL. (2008). Initiation of Wnt signaling: control of Wnt coreceptor Lrp6 (1994). Dishevelled is required during wingless signaling to establish phosphorylation/activation via frizzled, dishevelled and axin func- both cell polarity and cell identity. Development 120: 347–360. tions. Development 135: 367–375.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene